Dynamics of the Warsaw Stock Exchange index as analysed by the Mittag-Leffler function Marzena Kozłowska and Ryszard Kutner Department of Physics, Warsaw University Physics of Socio-Economic Systems AKSOE 2006 Dresden, March 26-31, 2006
Dynamics of the Warsaw Stock Exchange index as analysed by the fractional relaxation equation Marzena Kozłowska, Ryszard Kutner Division of Physics Education Institute of Experimental Physics Department of Physics, Warsaw University 2 Ogólnopolskie Sympozjum z Fizyki w Ekonomii i Naukach Społecznych Kraków, 21-22 kwiecień 2006
The authors are very grateful to Market Operations and Development of Warsaw Stock Exchange for supplied empirical data
Schedule Motivation and inspiration Model: fractional risings and relaxations decorated by oscillations Comparison with empirical data Microscopic model: hierarchical network of investors Conclusions
Motivation and inspiration WIG is the oldest index of the WSE. Duration of its peaks covers ~3/4 time-range. Definition of WIG analogous to S&P500. Our aim is to describe: Slowing down of risings and relaxations within intermediate (and possibly long) time-ranges of temporal maximums of WIG. Systematic oscillations of WIG within maximums. Possibly sudden increase of market activity in the vicinity of temporal maximums of WIG. Retarded feedback of WIG (a kind of memory): main effect responsible for temporal maximums.
Model: fractional risings, relaxations and oscillations. Model consists of two steps Linear, ordinary differential equation of the first order without any retardation describing evolution of a hypothetical auxiliary index. Glöckle-Nonnenmacher conjecture which transforms this equation to a more general fractional differential one which already desribes evolution of an empirical index. This was done in analogy to similar step performed for viscoelastic materials.
WIG dzienny na zamknięciu: 16.04.91-31.12.05
Definitions of indexes Defintion of WIG and S&P500 WIG(t), S&P500(t) ~ K t L j t X j t L j t K t : correction : current number of stocks of j-th company on stock market X j t : price of stock of j-th company
Model: fractional risings, relaxations and oscillations Instantaneous decomposition of WIG(t) WIG(t) = A U(t) + B V(t) ( 0) U(t): instantaneous offset between demand and supply V(t): volume trade A, B, C, D, E: coefficients Instantaneous dynamics dv t dt =C V t +D WIG t +E dwig t dt
Model: fractional risings, relaxations and oscillations Conveniet equation without V(t): dwig t = 1 dt τ WIG t ± A' τ 0 U t +A' du t dt A'= A 1 BE, τ 0 = 1 C, τ= C+BD 1 BE 1 Conjecture: usual differentiations are replaced by fractional differentiations
Model: fractional risings, relaxations and oscillations Dynamics of WIG is described by the linear fractional differential equation. Free case U=0 (for 0 < α 1): dwig t = 1 dt τ α 0 D 1 α t WIG t, 0 <α 1 0 D t 1 α WIG t = d dt 0 D t α WIG t 0 D α t WIG t = 1 Γ α WIG y Retarded feedback well seen t y 1 α dy
Free solution WIG t =WIG 0 E α t τ α, 0 <α 1 Mittag-Leffler fnction (generalised exponent) E α t τ α = t / τ α n Γ 1 +αn
Characteristic limits of Mittag-Leffler function Stretched exponential function or Kohlrausch-Williams-Watts α (KWW) decay:, 0 <α 1, t τ E α t τ α exp t τ Power-law or Nutting law: E α t τ α 1 Γ 1 α 1 t / τ, 0 <α 1, t τ α
Model: fractional risings, relaxations and oscillations Dynamics of WIG is described by the linear fractional differentia equation. Full case (for 0 < α 1): dwig t = 1 dt τ α 0 D 1 α t WIG t A' α τ 0 0 D 1 α t U t +A ' du t dt U(t) instantaneous offset between demand and supply. For example, a wiggle: U t ~ exp i ω Δω t exp i ω+δω t
Approximate solution α α WIG t t MAX ~ E t t MAX τ +C cos ω t t MAX cos Δω t t MAX Comparison with empirical data gives: 0.25 α 0.42 60 τ 320 0.1 ω 0.2 0.01 Δω 0.02 1500 C 4500
Daily closing price of WIG
Daily closing price of WIG20
Daily closing price of Elektrim-Vivendi Company
Temporal maximum A of WIG
Temporal maximum a of WIG
Temporal maximum B of WIG
Maximum B: KWW & N limits of Mittag-Leffler function
Maximum B: KWW & N limits of Mittag-Leffler function
Temporal maximum C of WIG
Model mikroskopowy: hierarchia inwestorów
Podsumowanie Praktycznie rzecz biorąc, w obrębie maksimów inwestorzy stanowią układy (sieci) pośrednie pomiędzy opisywanymi prawem KWW a prawem Nuttinga. Maksima dobrze opisuje się funkcjami Mittag-Lefflera udekorowanymi różnego rodzaju oscylacjami. Są one rozwiązaniem ułamkowych równań różniczkowych, za którymi stoi przede wszystkim efekt opóźnionego sprzężenia zwrotnego w dynamice indeksu. Równania te można otrzymać z poziomu mikro zakładając hierarchiczną strukturę powiązań pomiędzy inwestorami.
Podsumowanie c.d. dx t dt Teoretyczna szybkość rozbiega się jak 1 t t MAX 1 gdy t t MAX z lewej lub prawej strony. Przypuszczamy, że mamy tutaj do czynienia z dynamicznym przejściem fazowym (faza wzrostu faza spadku). Podejście można stosować do innych indeksów a także cen akcji.